Forecasting Weekly Mathematics Performance through ARIMA Modeling: An Action Research in Philippine Elementary Education

Background This action research investigates the application of AutoRegressive Integrated Moving Average (ARIMA) modeling to predict weekly mathematics performance among 10– to 11–year–old students in an elementary school in Cebu, Philippines. The study aims to identify trends in quiz scores to evaluate the effectiveness of instructional strategies applied throughout the quarter. Methods Data were collected from a heterogeneous class of boys and girls with varied learning profiles over a period of eighteen weeks. The researcher averaged the weekly quiz scores and analyzed them using the ARIMA(1,1,1) model in Gretl software to detect patterns and generate short-term performance forecasts. Results The results revealed that student performance fluctuated based on the complexity of the topic and the instructional method, with higher scores observed in weeks that emphasized strategy-intensive approaches. Forecasted data showed a stabilizing trend at around 78.5% in the subsequent weeks. Conclusion This study concluded that ARIMA modeling is a valuable tool for supporting data-informed, responsive teaching in the mathematics classroom. Fluctuations in student performance aligned with topic difficulty and instructional strategy, while forecasted scores indicated the positive impact of collaborative and strategy-focused methods. The study affirms the role of classroom-based forecasting in enhancing real-time instructional decision-making, encouraging continuous data collection and pedagogical reflection in basic education.